Skip to main content

Online participation and outcomes across student backgrounds

ECI 586 Final Project

Author

Ran Wu

1 Prepare

1.1 Research Purpose and Question

The purpose of this project is to examine how students’ socioeconomic and educational backgrounds relate to their online participation and final course outcomes in distance learning.

The research question is:

How do online participation and final course outcomes differ by students’ prior education and the socioeconomic deprivation of their residential areas?

The analysis focuses on one course offering and examines participation across the full observed period. An additional comparison uses a fixed eight-week window to examine whether background differences are already apparent early in the course. Comparisons between the two periods use the same students with recorded activity during the early window.

The findings are intended for distance education instructors and instructional designers, as well as student support staff. They may inform a review of academic guidance and identify areas where further feedback about students’ learning experiences would be useful.

1.2 Data Sources and Variables

The project uses the Open University Learning Analytics Dataset, which contains anonymized records from courses offered by the UK Open University during 2013-2014 (Kuzilek et al., 2017). Within this distance education setting, students accessed teaching materials through a virtual learning environment that recorded their interactions with course resources.

The analysis draws on two files. The student information file, studentInfo.csv, contains background characteristics and final course results. The activity file, studentVle.csv, provides daily interaction counts for individual learning resources. In general, these two sources provide information about student backgrounds, recorded participation and course outcomes.

The main variables are summarized below.

Variable Identifier Purpose
Area-level socioeconomic deprivation imd_band Defines groups based on students’ residential circumstances
Prior education highest_education Defines groups according to qualifications before entering the course
Final course outcome final_result Describes students’ results at the end of the course
Recorded interaction count sum_click Provides the counts used to calculate total interactions
Relative course date date Identifies activity dates for calculating active days and selecting the first eight weeks

Both files are first restricted to the same course offering. Then, activity records are summarized by student and joined to student information using id_student.
Lower imd_band scores indicate greater area-level deprivation, while higher scores indicate less deprivation.

2 Wrangle

2.1 Inspect Student Information

The student information file is inspected to identify its structure and the available course offerings. Student records are counted for each combination of course module and presentation.

library(tidyverse)
── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.2.1     ✔ readr     2.2.0
✔ forcats   1.0.1     ✔ stringr   1.6.0
✔ ggplot2   4.0.4     ✔ tibble    3.3.1
✔ lubridate 1.9.5     ✔ tidyr     1.3.2
✔ purrr     1.2.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
student_info <- read_csv("data/studentInfo.csv")
Rows: 32593 Columns: 12
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (9): code_module, code_presentation, gender, region, highest_education, ...
dbl (3): id_student, num_of_prev_attempts, studied_credits

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
glimpse(student_info)
Rows: 32,593
Columns: 12
$ code_module          <chr> "AAA", "AAA", "AAA", "AAA", "AAA", "AAA", "AAA", …
$ code_presentation    <chr> "2013J", "2013J", "2013J", "2013J", "2013J", "201…
$ id_student           <dbl> 11391, 28400, 30268, 31604, 32885, 38053, 45462, …
$ gender               <chr> "M", "F", "F", "F", "F", "M", "M", "F", "F", "M",…
$ region               <chr> "East Anglian Region", "Scotland", "North Western…
$ highest_education    <chr> "HE Qualification", "HE Qualification", "A Level …
$ imd_band             <chr> "90-100%", "20-30%", "30-40%", "50-60%", "50-60%"…
$ age_band             <chr> "55<=", "35-55", "35-55", "35-55", "0-35", "35-55…
$ num_of_prev_attempts <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
$ studied_credits      <dbl> 240, 60, 60, 60, 60, 60, 60, 120, 90, 60, 60, 60,…
$ disability           <chr> "N", "N", "Y", "N", "N", "N", "N", "N", "N", "N",…
$ final_result         <chr> "Pass", "Pass", "Withdrawn", "Pass", "Pass", "Pas…
course_counts <- student_info %>%
  count(code_module, code_presentation, sort = TRUE)

course_counts
# A tibble: 22 × 3
   code_module code_presentation     n
   <chr>       <chr>             <int>
 1 CCC         2014J              2498
 2 FFF         2014J              2365
 3 BBB         2014J              2292
 4 FFF         2013J              2283
 5 BBB         2013J              2237
 6 DDD         2013J              1938
 7 CCC         2014B              1936
 8 DDD         2014J              1803
 9 BBB         2013B              1767
10 FFF         2013B              1614
# ℹ 12 more rows

The summary shows 22 combinations of course modules and presentations. Each combination of module and presentation represents a specific course offering.

2.2 Select a Course Offering

A single course offering was selected to compare student groups. Record counts and missing information were examined for each module and presentation to identify a suitable sample. The deprivation check included both NA values and the “?” label, which represents unknown information.

course_check <- student_info %>%
  group_by(code_module, code_presentation) %>%
  summarize(student_records = n(),
            missing_deprivation = sum(is.na(imd_band) | imd_band == "?"),
            missing_education = sum(is.na(highest_education)),
            missing_outcome = sum(is.na(final_result))) %>%
  ungroup() %>%
  arrange(desc(student_records))
`summarise()` has regrouped the output.
ℹ Summaries were computed grouped by code_module and code_presentation.
ℹ Output is grouped by code_module.
ℹ Use `summarise(.groups = "drop_last")` to silence this message.
ℹ Use `summarise(.by = c(code_module, code_presentation))` for per-operation
  grouping (`?dplyr::dplyr_by`) instead.
course_check
# A tibble: 22 × 6
   code_module code_presentation student_records missing_deprivation
   <chr>       <chr>                       <int>               <int>
 1 CCC         2014J                        2498                 146
 2 FFF         2014J                        2365                  99
 3 BBB         2014J                        2292                  13
 4 FFF         2013J                        2283                 101
 5 BBB         2013J                        2237                  16
 6 DDD         2013J                        1938                  80
 7 CCC         2014B                        1936                 108
 8 DDD         2014J                        1803                  81
 9 BBB         2013B                        1767                  21
10 FFF         2013B                        1614                  95
# ℹ 12 more rows
# ℹ 2 more variables: missing_education <int>, missing_outcome <int>

The dataset contained 22 course offerings. CCC in 2014J contained the largest number of student records (2498). Deprivation information was recorded as “?” for 146 students in this offering. No NA values were detected in prior education or final course outcome.

CCC in 2014J was selected because it provided the largest number of student records.

selected_students <- student_info %>%
  filter(code_module == "CCC",
    code_presentation == "2014J")

selected_students %>%
  count(code_module, code_presentation)
# A tibble: 1 × 3
  code_module code_presentation     n
  <chr>       <chr>             <int>
1 CCC         2014J              2498

Student identifiers are checked for repeated records to ensure that each student contributes one background record within the selected offering.

selected_students %>%
  count(id_student) %>%
  filter(n > 1)
# A tibble: 0 × 2
# ℹ 2 variables: id_student <dbl>, n <int>

No repeated student identifiers were found in the selected offering.

2.3 Inspect Activity Data

The student activity file contains the interaction records needed to measure online participation. Records were filtered to CCC in 2014J so that the activity data correspond to the selected student information.

student_vle <- read_csv("data/studentVle.csv")
Rows: 10655280 Columns: 6
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (2): code_module, code_presentation
dbl (4): id_student, id_site, date, sum_click

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
selected_vle <- student_vle %>%
  filter(code_module == "CCC",
    code_presentation == "2014J")

glimpse(selected_vle)
Rows: 711,646
Columns: 6
$ code_module       <chr> "CCC", "CCC", "CCC", "CCC", "CCC", "CCC", "CCC", "CC…
$ code_presentation <chr> "2014J", "2014J", "2014J", "2014J", "2014J", "2014J"…
$ id_student        <dbl> 582530, 582530, 582530, 582530, 582530, 582530, 5825…
$ id_site           <dbl> 909232, 909230, 909071, 909231, 909013, 909260, 9092…
$ date              <dbl> -18, -18, -18, -18, -18, -18, -18, -18, -18, -18, -1…
$ sum_click         <dbl> 5, 1, 7, 1, 10, 2, 2, 4, 1, 4, 1, 1, 1, 1, 1, 14, 1,…

There are 711646 activity records for CCC in 2014J. The selected data contain six variables, including code_module, code_presentation, id_student, id_site, date, and sum_click.

2.4 Create Student Participation Indicators

Student participation indicators are created to connect activity records with student backgrounds and course outcomes. Total clicks are calculated by adding each student’s recorded clicks. Active days are calculated by counting distinct dates with activity records. Full-period participation is measured using all recorded activity dates in the selected offering.

student_participation <- selected_vle %>%
  group_by(id_student) %>%
  summarize(total_clicks = sum(sum_click),
    active_days = n_distinct(date),
    .groups = "drop")

student_participation
# A tibble: 2,302 × 3
   id_student total_clicks active_days
        <dbl>        <dbl>       <int>
 1      23698          910          70
 2      25261          997          39
 3      27116         1847         126
 4      28787          645          59
 5      28952          105          12
 6      29411          329          19
 7      29420          816          49
 8      29639         6986         245
 9      30096          714          54
10      30850         2258         130
# ℹ 2,292 more rows

There are 2302 student records with three variables: id_student, total_clicks, and active_days.

2.5 Create Early Participation Indicators

Participation is also indicated during the first eight weeks to reduce the influence of differences in how long students remain enrolled. The fixed window covers days 0-55 from the official course start. Early clicks and active days are calculated using the same aggregation methods mentioned early.

early_participation <- selected_vle %>%
  filter(date >= 0, date < 56) %>%
  group_by(id_student) %>%
  summarize(early_clicks = sum(sum_click),
            early_active_days = n_distinct(date), 
            .groups = "drop")

early_participation
# A tibble: 2,208 × 3
   id_student early_clicks early_active_days
        <dbl>        <dbl>             <int>
 1      23698          211                21
 2      25261          794                27
 3      27116          519                30
 4      28787          120                16
 5      28952           77                11
 6      29411          182                14
 7      29420          150                13
 8      29639         2998                48
 9      30096          199                13
10      30850          779                25
# ℹ 2,198 more rows

There are 2208 student records.

2.6 Merge Data and Select the Sample

The sample for the full-period analysis is created by joining student information with participation summaries using id_student. An inner join retains students with matching activity records. Records where imd_band equals “?” are excluded because these students cannot be assigned to a deprivation group.

analysis_data <- selected_students %>%
  inner_join(student_participation, by = "id_student") %>%
  filter(imd_band != "?")

analysis_data %>%
  summarize(full_period_students = n())
# A tibble: 1 × 1
  full_period_students
                 <int>
1                 2162

The full-period sample contained 2162 students with recorded platform activity and identifiable deprivation information. Overall, 336 records were excluded from the original 2498 student records.

Then, a separate comparison sample is created from this dataset by retaining students with recorded activity during the first eight weeks. The early participation indicators are added to their existing records, so both observation periods can be examined using the same students.

analysis_data_early <- analysis_data %>%
  inner_join(early_participation, by = "id_student")

analysis_data_early %>%
  summarize(early_window_students = n())
# A tibble: 1 × 1
  early_window_students
                  <int>
1                  2070

This additional join retained 2070 students and excluded 92 without recorded activity during the early window.

The dataset analysis_data is used for the full-period analyses, while analysis_data_early is used for comparisons between the two observation periods.

3 Analyze

3.1 Describe Student Backgrounds

Counts and percentages show the representation of each deprivation band and prior education category.

deprivation_counts <- analysis_data %>%
  count(imd_band) %>%
  mutate(percentage = n / sum(n) * 100)

deprivation_counts
# A tibble: 10 × 3
   imd_band     n percentage
   <chr>    <int>      <dbl>
 1 0-10%      182       8.42
 2 10-20      199       9.20
 3 20-30%     233      10.8 
 4 30-40%     239      11.1 
 5 40-50%     221      10.2 
 6 50-60%     226      10.5 
 7 60-70%     220      10.2 
 8 70-80%     217      10.0 
 9 80-90%     226      10.5 
10 90-100%    199       9.20

Students were distributed relatively evenly across the ten deprivation bands, with each band accounting for 8.4%-11.1% of the full-period sample. The 30-40% band contained the largest group (239 students, 11.1%), while the 0-10% band contained the smallest group (182 students, 8.4%).

education_counts <- analysis_data %>%
  count(highest_education, sort = TRUE) %>%
  mutate(percentage = n / sum(n) * 100)

education_counts
# A tibble: 5 × 3
  highest_education               n percentage
  <chr>                       <int>      <dbl>
1 A Level or Equivalent        1060     49.0  
2 Lower Than A Level            599     27.7  
3 HE Qualification              454     21.0  
4 Post Graduate Qualification    31      1.43 
5 No Formal quals                18      0.833

Prior education was more unevenly distributed. Almost half of the students held A Level or Equivalent qualifications (1060 students, 49.0%). Lower Than A Level accounted for 599 students (27.7%), followed by HE Qualification with 454 students (21.0%). The remaining categories were considerably smaller: Post Graduate Qualification contained 31 students (1.4%), and No Formal quals contained 18 students (0.8%).

3.2 Describe Online Participation Distributions

Histograms display the distributions of total clicks and active days in the full-period sample.

3.2.1 Total Clicks

ggplot(analysis_data, aes(x = total_clicks)) +
  geom_histogram(bins = 25,
                 fill = "cadetblue",
                 color = "white") +
  labs(title = "Distribution of Total Clicks",
       x = "Total clicks",
       y = "Number of students")

Total clicks were strongly right-skewed. Most students recorded relatively low click counts, while a small number of highly active students formed a long right tail.

3.2.2 Active Days

ggplot(analysis_data, aes(x = active_days)) +
  geom_histogram(bins = 25,
                 fill = "darkorange",
                 color = "white") +
  labs(title = "Distribution of Active Days",
       x = "Active days",
       y = "Number of students")

Active days also showed a right-skewed distribution. Most students had fewer than 100 active days, while relatively few recorded activity on more than 200 days.

3.3 Compare Full-Period Participation Across Deprivation Bands

Total clicks and active days are summarized by deprivation band to examine differences in online participation. Means and medians describe participation levels within each group. Standard deviations show how much students’ participation varies within the same band.

deprivation_participation <- analysis_data %>%
  group_by(imd_band) %>%
  summarize(student_records = n(),
            mean_clicks = mean(total_clicks),
            median_clicks = median(total_clicks),
            sd_clicks = sd(total_clicks), 
            mean_active_days = mean(active_days),
            median_active_days = median(active_days),
            sd_active_days = sd(active_days),
            .groups = "drop")

deprivation_participation
# A tibble: 10 × 8
   imd_band student_records mean_clicks median_clicks sd_clicks mean_active_days
   <chr>              <int>       <dbl>         <dbl>     <dbl>            <dbl>
 1 0-10%                182       1209.          529      1660.             55.0
 2 10-20                199       1030.          534      1302.             50.5
 3 20-30%               233       1073.          549      1653.             50.5
 4 30-40%               239       1213.          650      1790.             53.9
 5 40-50%               221       1040.          599      1386.             50.1
 6 50-60%               226       1218.          752.     1533.             56.6
 7 60-70%               220       1133.          600.     1414.             56.8
 8 70-80%               217       1130.          688      1246.             56.5
 9 80-90%               226       1485.          879      2186.             66.0
10 90-100%              199       1227.          808      1286.             60.1
# ℹ 2 more variables: median_active_days <dbl>, sd_active_days <dbl>

The 80-90% band showed the highest participation, with a mean of 1484.8 clicks and 66.0 active days. Its medians were also the highest, at 879 clicks and 57.5 active days. Mean clicks across the ten bands ranged from 1029.9 to 1484.8, while mean active days ranged from 50.1 to 66.0. Participation did not increase consistently across successive bands.

Within-group variation was substantial. The 80-90% band had the largest standard deviations for both clicks (2186.0) and active days (57.4). Means exceeded medians in every band, suggesting that students with particularly high activity raised the group averages.

Boxplots compare participation distributions across deprivation bands. They display group medians, within-group variation, and potential outliers.

ggplot(analysis_data,
       aes(x = total_clicks, y = imd_band)) +
  geom_boxplot(fill = "dodgerblue3") +
  labs(title = "Total Clicks by Deprivation Band",
       x = "Total clicks",
       y = "Deprivation band")

ggplot(analysis_data,
       aes(x = active_days, y = imd_band)) +
  geom_boxplot(fill = "indianred") +
  labs(
    title = "Active Days by Deprivation Band",
    x = "Active days",
    y = "Deprivation band"
  )

Both indicators showed substantial overlap across deprivation bands, with considerable within-group variation and high-value outliers.

3.4 Compare Full-Period Participation Across Prior Education Levels

The same descriptive measures are used as in the deprivation comparison.

education_participation <- analysis_data %>%
  group_by(highest_education) %>%
  summarize(student_records = n(),
            mean_clicks = mean(total_clicks),
            median_clicks = median(total_clicks),
            sd_clicks = sd(total_clicks),
            mean_active_days = mean(active_days),
            median_active_days = median(active_days),
            sd_active_days = sd(active_days),
            .groups = "drop")

education_participation
# A tibble: 5 × 8
  highest_education          student_records mean_clicks median_clicks sd_clicks
  <chr>                                <int>       <dbl>         <dbl>     <dbl>
1 A Level or Equivalent                 1060       1075.          640.     1366.
2 HE Qualification                       454       1435.          832.     1692.
3 Lower Than A Level                     599       1145.          524      1823.
4 No Formal quals                         18        776.          612.      962.
5 Post Graduate Qualificati…              31       1730.         1330      1454.
# ℹ 3 more variables: mean_active_days <dbl>, median_active_days <dbl>,
#   sd_active_days <dbl>

Students with Post Graduate Qualifications showed the highest participation, with means of 1730.1 clicks and 77.4 active days. Their medians were also the highest, at 1330 clicks and 60 days. HE Qualification ranked second on both measures, with medians of 831.5 clicks and 51 active days.

Lower Than A Level had the lowest median clicks (524). Its median active days was 32, matching No Formal quals. Click counts varied substantially within educational groups, with the largest standard deviation in Lower Than A Level (1823.4).

Boxplots are also used to examine participation distributions within each prior education category.

ggplot(analysis_data,
       aes(x = total_clicks, y = highest_education)) +
  geom_boxplot(fill = "cadetblue") +
  labs(title = "Total Clicks by Education",
       x = "Total clicks",
       y = "Prior education") +
  theme( plot.title = element_text(size = 14), axis.text.y = element_text(size = 10))

ggplot(analysis_data,
       aes(x = active_days, y = highest_education)) +
  geom_boxplot(fill = "darkorange") +
  labs(title = "Active Days by Education",
       x = "Active days",
       y = "Prior education") +
  theme(plot.title = element_text(size = 14),
        axis.text.y = element_text(size = 10))

Both indicators showed overlapping distributions across education categories, with higher medians in the postgraduate group and high-value outliers in several other groups.

3.5 Compare Early and Full-Period Participation Across Deprivation Bands

Early and full-period participation are compared to examine whether differences across deprivation bands are already apparent during the first eight weeks. Median clicks and active days are summarized for both observation periods.

early_deprivation_comparison <- analysis_data_early %>%
  group_by(imd_band) %>%
  summarize(student_records = n(),
            median_full_clicks = median(total_clicks),
            median_early_clicks = median(early_clicks),
            median_full_days = median(active_days),
            median_early_days = median(early_active_days),
            .groups = "drop")

early_deprivation_comparison
# A tibble: 10 × 6
   imd_band student_records median_full_clicks median_early_clicks
   <chr>              <int>              <dbl>               <dbl>
 1 0-10%                172               559                 250.
 2 10-20                186               622.                259 
 3 20-30%               225               578                 242 
 4 30-40%               227               709                 268 
 5 40-50%               212               644.                208.
 6 50-60%               216               784                 290.
 7 60-70%               213               641                 241 
 8 70-80%               210               720.                267 
 9 80-90%               215               944                 286 
10 90-100%              194               876                 290.
# ℹ 2 more variables: median_full_days <dbl>, median_early_days <dbl>

Median early active days ranged from 14 to 17, with the highest values in the 90-100% band (17 days) and the 80-90% band (16 days). Median early clicks were highest in the 50-60% and 90-100% bands, both at 290.5 clicks. In comparison, the 80-90% band had the highest full-period median clicks (944). Therefore, the ranking of deprivation bands varied between the two observation periods.

Point plots compare median participation across the early and full observation periods. Different colors distinguish the periods and help show whether group rankings are similar.

ggplot(early_deprivation_comparison, aes(y = imd_band)) +
  geom_point(aes(x = median_early_clicks, 
                 color = "First 8 weeks"),
             size = 3) +
  geom_point(aes(x = median_full_clicks, color = "Full period"),
             size = 3) +
  labs(title = "Median Clicks by Deprivation Band",
       x = "Median total clicks",
       y = "Deprivation band",
       color = "Period") +
  theme(legend.position = "bottom")

ggplot(early_deprivation_comparison, aes(y = imd_band)) +
  geom_point(
    aes(x = median_early_days, color = "First 8 weeks"),
    size = 3) +
  geom_point(aes(x = median_full_days,
                 color = "Full period"),
    size = 3
  ) +
  labs(
    title = "Median Active Days by Deprivation Band",
    x = "Median active days",
    y = "Deprivation band",
    color = "Period"
  ) +
  theme(legend.position = "bottom")

According to the plots, the 80-90% and 90-100% bands generally ranked high in both periods.

3.6 Compare Early and Full-Period Participation Across Prior Education Levels

The same comparison is applied to prior education categories to examine whether the participation patterns observed across the full period are also present early in the course.

early_education_comparison <- analysis_data_early %>%
  group_by(highest_education) %>%
  summarize(student_records = n(),
            median_full_clicks = median(total_clicks),
            median_early_clicks = median(early_clicks),
            median_full_days = median(active_days),
            median_early_days = median(early_active_days),
            .groups = "drop"
  )

early_education_comparison
# A tibble: 5 × 6
  highest_education       student_records median_full_clicks median_early_clicks
  <chr>                             <int>              <dbl>               <dbl>
1 A Level or Equivalent              1013                679                255 
2 HE Qualification                    444                882                306.
3 Lower Than A Level                  565                576                251 
4 No Formal quals                      17                640                234 
5 Post Graduate Qualific…              31               1330                348 
# ℹ 2 more variables: median_full_days <dbl>, median_early_days <dbl>

The postgraduate group had the highest median early participation, with 348 clicks and 19 active days. Students with HE Qualifications ranked second, with 305.5 clicks and 16 active days. Both groups also ranked highest on the full-period indicators. Their relatively high participation was already visible during the first eight weeks.

The same point plots are applied to prior education categories.

ggplot(early_education_comparison, aes(y = highest_education)) +
  geom_point(aes(x = median_early_clicks, 
                 color = "First 8 weeks"),
             size = 2) +
  geom_point(
    aes(x = median_full_clicks, color = "Full period"),
    size = 2) +
  labs(title = "Median Clicks by Education",
       x = "Median total clicks",
       y = "Prior education",
       color = "Period") +
  theme(legend.position = "bottom")

ggplot(early_education_comparison, aes(y = highest_education)) +
  geom_point(aes(x = median_early_days, 
                 color = "First 8 weeks"),
             size = 2) +
  geom_point(aes(x = median_full_days, 
                 color = "Full period"),
    size = 2) +
  labs(title = "Median Active Days by Education",
       x = "Median active days",
       y = "Prior education",
       color = "Period") +
  theme(legend.position = "bottom")

3.7 Compare Course Outcomes Across Deprivation Bands

Final course outcomes are compared across deprivation bands to examine differences in students’ course results. Percentages are calculated within each band so that outcome distributions can be compared across groups of different sizes.

deprivation_outcomes <- analysis_data %>%
  count(imd_band, final_result) %>%
  group_by(imd_band) %>%
  mutate(percentage = n / sum(n) * 100) %>%
  ungroup()

deprivation_outcomes
# A tibble: 40 × 4
   imd_band final_result     n percentage
   <chr>    <chr>        <int>      <dbl>
 1 0-10%    Distinction     16       8.79
 2 0-10%    Fail            29      15.9 
 3 0-10%    Pass            52      28.6 
 4 0-10%    Withdrawn       85      46.7 
 5 10-20    Distinction     18       9.05
 6 10-20    Fail            44      22.1 
 7 10-20    Pass            51      25.6 
 8 10-20    Withdrawn       86      43.2 
 9 20-30%   Distinction     26      11.2 
10 20-30%   Fail            41      17.6 
# ℹ 30 more rows

Withdrawal was the most common outcome in eight of the ten deprivation bands. The withdrawal percentage was highest in the 20-30% band (48.5%) and lowest in the 80-90% band (30.5%).

The 80-90% band had the highest Pass percentage (35.0%), while the 90-100% band had the highest Distinction percentage (18.1%). The 0–10% band had the lowest Distinction percentage (8.8%).

A horizontal percentage stacked bar chart displays the composition of course outcomes within each deprivation band.

ggplot(deprivation_outcomes,
       aes(x = percentage,
           y = imd_band,
           fill = final_result)) +
  geom_col() +
  labs(title = "Course Outcomes by Deprivation Band",
       x = "Students (%)",
       y = "Deprivation band",
       fill = "Course outcome")

3.8 Compare Course Outcomes Across Prior Education Levels

The same procedure is applied to prior education categories to examine how course outcome distributions differ across educational backgrounds.

education_outcomes <- analysis_data %>%
  count(highest_education, final_result) %>%
  group_by(highest_education) %>%
  mutate(percentage = n / sum(n) * 100) %>%
  ungroup()

education_outcomes
# A tibble: 20 × 4
   highest_education           final_result     n percentage
   <chr>                       <chr>        <int>      <dbl>
 1 A Level or Equivalent       Distinction    133      12.5 
 2 A Level or Equivalent       Fail           180      17.0 
 3 A Level or Equivalent       Pass           345      32.5 
 4 A Level or Equivalent       Withdrawn      402      37.9 
 5 HE Qualification            Distinction     95      20.9 
 6 HE Qualification            Fail            73      16.1 
 7 HE Qualification            Pass           143      31.5 
 8 HE Qualification            Withdrawn      143      31.5 
 9 Lower Than A Level          Distinction     26       4.34
10 Lower Than A Level          Fail           122      20.4 
11 Lower Than A Level          Pass           156      26.0 
12 Lower Than A Level          Withdrawn      295      49.2 
13 No Formal quals             Distinction      2      11.1 
14 No Formal quals             Fail             1       5.56
15 No Formal quals             Pass             5      27.8 
16 No Formal quals             Withdrawn       10      55.6 
17 Post Graduate Qualification Distinction     12      38.7 
18 Post Graduate Qualification Fail             6      19.4 
19 Post Graduate Qualification Pass             8      25.8 
20 Post Graduate Qualification Withdrawn        5      16.1 

The postgraduate group had the highest Distinction percentage (38.7%, 12 of 31 students) and the lowest withdrawal percentage (16.1%). Withdrawal was highest among students with No Formal quals (55.6%, 10 of 18 students), followed by Lower Than A Level (49.2%). Lower Than A Level also had the lowest Distinction percentage (4.3%), compared with 20.9% among students with HE Qualifications.

A horizontal percentage stacked bar chart displays course outcome distributions across prior education categories.

ggplot(education_outcomes,
       aes(x = percentage,
           y = highest_education,
           fill = final_result)) +
  geom_col() +
  labs(title = "Course Outcomes by Education",
       x = "Students (%)",
       y = "Prior education",
       fill = "Outcome") +
  theme(plot.title = element_text(size = 12),
        axis.text.y = element_text(size = 8),
        legend.position = "bottom")

4 Communication

4.1 Findings

This project suggests that online participation and course outcomes in the selected course offering were linked to students’ educational and socioeconomic backgrounds.

For prior education, the postgraduate and HE Qualification groups showed higher median participation. The postgraduate group also had the highest Distinction percentage and the lowest withdrawal percentage. In contrast, withdrawal was particularly common among students with No Formal quals and Lower Than A Level. The two observation periods comparison strengthened this finding, as the postgraduate and HE Qualification groups ranked highest on both participation indicators in both observation periods. Therefore, this suggests that support needs may emerge early in the course. Early activity records could provide a starting point for conversations about students’ learning circumstances.

For area-level deprivation, the 80-90% and 90-100% bands generally combined higher participation with more favorable course outcomes. Withdrawal ranged from 48.5% in the 20-30% band to 30.5% in the 80-90% band. However, unlike prior education, the participation rankings changed between the two observation periods, so the pattern depended partly on when activity was examined. This suggests that area-level deprivation is associated with outcomes but is a weaker and less stable indicator than prior education.

4.2 Potential Actions

These findings point to several actions for instructors, course designers, and student support teams. First, participation gaps were already visible within the first eight weeks. Instructors could check participation data during the early weeks of the course and follow up with students who show low activity. Second, students with No Formal quals or Lower Than A Level qualifications withdrew at the highest rates. Support could include an orientation to online study, additional study resources, and regular check-ins. Third, the deprivation pattern was less stable and varied widely within bands, so it should not be used on its own to flag students. Finally, interviews or surveys with students who withdrew could help explore student’ reasons for withdrawal and their experiences of course support.

4.3 Limitations and Ethical Considerations

This project has several limitations. First, it covers only one course offering, so the results may not generalize to other courses or years. Second, participation was measured through click counts, which capture online activity but not the quality of learning or any study that took place offline. Third, the analysis is descriptive. It shows associations between student background and course outcomes but cannot establish causes, and it did not control for other factors such as age, workload, or disability. In addition, the deprivation results depended partly on the chosen observation periods, and some students had missing deprivation band values.

Ethically, the project uses a publicly available dataset released in anonymized form. However, deprivation band describes the area where a student lives rather than the student, so applying area-level patterns to individuals risks inaccurate assumptions. More broadly, these findings should not be used to label students or lower expectations for them. The purpose is to direct support where it may be

5 References

Kuzilek, J., Hlosta, M., & Zdrahal, Z. (2017). Open University Learning Analytics dataset. Scientific Data, 4, 170171. https://doi.org/10.1038/sdata.2017.171